Just a quick post today about the top ten NBA Playoff performances of 2013. To qualify, a player needs to have logged one of the top 100 scoring games in the postseason. This set the bar at 23 points in a game, which seems fair. There are some good reasons for making scoring the threshold, (like the fact that scoring is the only complete box-score stat we have) but that wasn't necessarily my reasoning. It was just easier that way.
Using Basketball Reference's Player Game Finder, I selected 100 games that fit the criteria. I then needed a linear box-score based metric that could assign weights to different stats, to give a one-number rating. I used Alternate Win Score (link for explanation at bottom) for the job. To the Wages of Wins crowd, let me explain.
-Alternate Win Score (AWS) gives an almost identical win-correlation to Win Score (WS), while significantly outperforming it, and all other linear box-score metrics, in prediction.
-Win Score (WS) needs a position adjustment, and Basketball Reference only sorts players into Guard/Wing/Forward categories. AWS doesn't need such an adjustment, making it the ideal metric to use.
To try and judge these performances on even ground, I adjusted for a few things. I made every score per-48 minutes, equalized pace to 100 possessions, and accounted for opposing team strength. This was done by calculating the AWS each team allowed on average during the regular season, and subtracting that from the AWS of a player's playoff performance.
Using Basketball Reference's Player Game Finder, I selected 100 games that fit the criteria. I then needed a linear box-score based metric that could assign weights to different stats, to give a one-number rating. I used Alternate Win Score (link for explanation at bottom) for the job. To the Wages of Wins crowd, let me explain.
-Alternate Win Score (AWS) gives an almost identical win-correlation to Win Score (WS), while significantly outperforming it, and all other linear box-score metrics, in prediction.
-Win Score (WS) needs a position adjustment, and Basketball Reference only sorts players into Guard/Wing/Forward categories. AWS doesn't need such an adjustment, making it the ideal metric to use.
To try and judge these performances on even ground, I adjusted for a few things. I made every score per-48 minutes, equalized pace to 100 possessions, and accounted for opposing team strength. This was done by calculating the AWS each team allowed on average during the regular season, and subtracting that from the AWS of a player's playoff performance.
That's all for now. I am kind of tempted to apply this method to each player's overall playoff stats. It would be interesting to see how we viewed playoff performance when pace and opponent strength was taken into account. The Alternate Win Score (AWS) link is below.
http://www.basketballprospectus.com/article.php?articleid=1985
http://www.basketballprospectus.com/article.php?articleid=1985